The Site With the Lamp

May 3, 2004

Sadly my quest to come up with a speech recognition system that runs on linux has come to nought. Whenever you google you find tons of pages, most of them refer to viavoice linux edition - a product that was discontinued two years ago. Closer inspection reveals that most of these pages were in fact created before 2002. Does that mean no one has even written about speech on linux let alone write a speech recognition system for linux?

More digging reveals things like CMU sphinx, (which seems to have been abondoned in 2002). You run into the occaisional nugget where people have tried to install the viavoice windows version or dragon naturally speaking on wine, usually without much success.

Unfortunately things like voice recognition are not things that can be easily built by independent open source developers. That's because there is a lot of tedius work involved and open source developers usually run away from it. (linux kernel and glibc probably being the exceptions).

It is not all bad news; you do find some Text To Speech (TTS) engines for the linux platform, hopefully one of these developers will initiate a recognition project and hopefully other developers will join them , i certainly wouldn't mind putting my shoulder to the wheel either.